# Changelog of PDF to Markdown & RAG Chunks: Word, Excel, PowerPoint (`quarrywise/pdf-to-markdown-rag`) Actor

- **URL**: https://apify.com/quarrywise/pdf-to-markdown-rag/changelog.md
- **Full Actor documentation**: https://apify.com/quarrywise/pdf-to-markdown-rag.md

## Changelog

### 0.1 (2026-10-26)

First release.

- PDF (born-digital), Word (.docx), PowerPoint (.pptx), Excel (.xlsx, .xls), HTML, Markdown, plain
  text and CSV to Markdown, JSON blocks or RAG chunks.
- PDF structure: heading levels from bookmarks, numbering and font size; ruled tables, simple
  aligned tables and tables continued on the next page; two-column reading order; running
  headers, footers and page numbers removed.
- Per-page quality report with text found, garble score, tables, images and warnings. Scanned
  pages are reported as needing OCR and are not charged.
- RAG chunks with heading path, page range, token estimate, chunk size and overlap.
- Inputs: links (public or signed), uploads, key-value store records and base64 files.
- Pay per event: per document and per page converted; failures and pages needing OCR are free.
- Robustness: every document is converted in a child process with a memory cap and a 10-minute
  limit, so a decompression bomb or a very heavy file costs one free error row, never the run.
- Headers and footers: only lines that repeat with a running page number or unchanged text are
  removed; lines with amounts, account numbers or other changing values always stay, and the
  removed texts are listed in the quality report.
- Chunks stay within the chunk size for long code blocks, list items and merged-cell tables.
- Large documents: Markdown or JSON blocks too large for one dataset row are saved as files.